Lecture 8b: Intervention Effect

Linguistic Theory and the Japanese Language INTERVENTION EFFECTS ARE SYNTACTIC Intervention effects (e.g., Beck 1996, Beck and Kim 1997, Hoji 1985, Rizzi 1992, Takahashi 1990). (1) *Was glaubt niemand, wen Karl gesehen whom Karl seen what believes nobody ‘Who does nobody believe that Karl saw?’ hat? has Intervenors in Japanese and Korean (2) Hoji 1985 (cf. also Hagstrom 1998) no? katta nani-o a. ?*Daremo-ga everyone-NOM what-ACC bought Q ‘What did everyone buy?’ universal quantifier no? katta nani-o b. ??Dareka-ga someone-NOM what-ACC bought Q ‘What did someone buy?’ existential quantifier c. *[John-ka Mary]-ga nani-o katta no? disjunction Mary]-NOM what-ACC bought Q [John-or ‘What did John or Mary buy?’ (3) Nani-oi daremo-ga/dareka-ga/[John-ka Mary]-ga ti katta no? what-ACCi everyone-NOM/someone-NOM/[John-or Mary]-NOM ti bought Q ‘What did everyone/someone/John or Mary buy?’ (4) Takahashi (1990) (cf.also Tanaka 1997) *Taroo-sika nani-o kaw-anakat-ta no? Taro-except what-ACC buy-NEG-PAST Q ‘What did only Taro buy?’ 1 (5) Quantifier-Induced Barrier (Beck 1996, Beck and Kim 1997) The first node that dominates a quantifer, its restriction, and its nuclear scope is a QUIB. * βi QUIB Q (e.g., Neg) ti Beck: only applies to LF movement (cf. also Pesetsky 2000) An Alternative, Pragmatics Approach (Tomioka 2004) Tomioka (2004) gives arguments that ostensibly favor a pragmatic approach to intervention. He notes that the intervenors identified in Japanese and Korean do not constitute a semantically coherent set. Instead, he observes that what they all have in common is their inability to topicalize —they do not occur with the topic marker –wa in Japanese or –(n)un in Korean. (6) “Anti-Topic Items” *daremo-wa/*dareka-wa/*[John-ka Mary]-wa/*Taro-sika-wa… everyone-TOP/someone-TOP/[John-or Mary]-TOP/Taro-except-TOP (7) The basic idea is that in the marginal wh-questions in (2) and (4), the ATIs, which occur on the left edge of the sentence, are in a position that is most natural for a topic to occur. (8) Taroo-wa nani-o katta no? Taro-TOP what-ACC bought Q ‘What did Taro buy?’ Intervention Effects May Be Cancelled in Certain Contexts (9) Embedding “embedded subjects are not topic marked” (p. 12) (cf. relevant discussion in Kuno 1973; Kuroda 1965, 1992) nani-o yonda to] omotteiru no? Kimi-wa [CP daremo-ga you-TOP [CP everyone-NOM what-ACC read- COMP think Q ‘What do you think that everyone read?’ 2 (10) Non-subject ATI Ken-wa daremo-ni nani-o ageta Ken-TOP everyone-DAT what-ACC gave ‘What did Ken give everyone?’ no? Q (11) Our (tentative) point: In example where the intervention effect is “cancelled,” the wh-phrase does not undergo (normal) movement. The wh-phrase in Tomioka’s examples are Dlinked in the sense of Pesetsky (1987). Problem with Tomioka’s Analysis “Intervention Effects are not Pragmatic,” Shigeru Miyagawa and Yoshio Endo, MIT and Yokohama National University ms., 2004. Daremo ‘everyone’ by itself may be interpreted as specific (cf. Hoji 1985). Adding “almost” avoids this non-quantificational interpretation (cf. Beck 1996, etc.). (12) a. Kimi-wa [CP hotondo daremo-ga Hanako-no-ronbun-o you-TOP [CP almost everyone-NOM Hanako-GEN-article-ACC yonda to] omotteiru no? read- COMP think Q ‘Do you think that almost everyone read Hanako’s article?’ b. ?*/OKKimi-wa [CP hotondo daremo-ga nani-o you-TOP [CP almost everyone-NOM what-ACC yonda to] omotteiru no? read- COMP think Q ATIs that do not allow this specific/non-specific alternative always induce intervention. (13) Non-subject ATI */OKKen-wa Hanako-ni-sika nani-o age-nakat-ta no? What did Ken-TOP everyone-DAT-except what-ACC gave-NEG-PAST Q ‘Ken give only Hanako?’ (14) Embedding nani-o kaw-anak-ta to] omotteiru no? */OKKimi-wa [CP Hanako-sika you-TOP [CP Hanako-except what-ACC buy- NEG-P COMP think Q ‘What do you think that only Hanako bought?’ Non-D-linked Environments: Tomioka’s examples turn out to be marginal. (15). [cp Hodondo daremo-ga nani-o yonda everyone-NOM what-ACC read [cp almost 3 koto]-ga mondai na no? fact]-NOM problem Q ‘What is the problem that almost everyone read?’ (16) ?*[cp Hotondo daremo-ga ittai nani-o yonda koto]-ga mondai na no? fact]-NOM problem Q [cp almost everyone-NOM world what-ACC read ‘What in the world is the problem that almost everyone read?’ (17) Tokorode, by the way *[cp Hotondo daremo-ga nani-o yonda koto]-ga mondai na no? [cp almost everyone-NOM what-ACC read fact]-NOM problem Q ‘By the way, what is the problem that almost everyone read?’ D-linking can cancel even the most basic intervention effect. (18) John-to Henry-to Mike-no uti, John-and Henry-and Mike-and among ?(Hotondo) daremo-ga dare-o kirat-te-iru no? (almost) everyone-NOM who-ACC hate Q ‘Among John, Henry, and Mike, who does almost everyone hate?’ (19) D-linking and superiority (Pesetsky 1987) a. *What did who read? b. Which book did which boy read? Why does D-linking “cancel” intervention effects? We can reduce it to the fact that when the wh-phrase is scrambled, this, too, cancels intervention effects. (20) Nani-oi daremo-ga/dareka-ga/[John-ka Mary]-ga ti katta no? what-ACCi everyone-NOM/someone-NOM/[John-or Mary]-NOM ti bought Q ‘What did everyone/someone/John or Mary buy?’ (21) Miyagawa (2004), based on observations by Beck (1996), Cresti (1995), Lahiri (2002). The restriction of a presuppositional wh-chain is interpreted “high” in the structure, in the vicinity of the wh quantification; the restriction of a non-presuppositional chain is interpreted “low” in the structure. (22) How many people do you wonder whether to invite? (only presuppositional) (23) D-linked wh-chains are presuppositional by nature. Hence the quantificational structure of a presuppositional wh-chain is identical to the scrambled case in (20). a. Overt form of a D-linked wh-question with QUIB: QUIB wh-phraseD-linked Verb. b. Quantificational form of (a): 4 Wh + Restrictioni QUIB ti Verb (24) Non-presuppositional wh-phrase and intervention effect: * wh x QUIB [x Restriction] Verb (25) Superiority: QUIB effect? a. *What did who read? b. Which book did which boy read? (26) a. What do you wonder whether to read? b. What the hell do you wonder whether to read? (bad??? All right???) Another problem: naze ‘why’ does not show intervention effect (Miyagawa 1997, etc) (27) a. Hanako-sika naze erabarenakatta Hanako-except why was.chosen. ‘Why was only Hanako chosen?’ no? Q b. Hotondo daremo-ga naze syukudai-o sinakatta almost everyone-NOM why homework-ACC didn’t.do ‘Why did almost everyone do the homework?’ (28)a. *Taroo-sika [CP Hanako-ga naze erabareta to] Taro-except [CP Hanako-NOM why was.chosen C] ‘Why does only Taro think that Hanako was chosen?’ no? Q omotte-inai no? think-NEG Q b. ?*Hotondo daremo-ga [CP Hanako-ga naze sigoto-o yameru to] almost everyone-NOM [CP Hanako-NOM why job-ACC quit C] omotteiru no? think Q ‘Why does almost everyone think that Hanako will quit her job?’ 5